Pakistan: Satellite Based Crop Monitoring and Forecasting ... · Development of human resource and...
Transcript of Pakistan: Satellite Based Crop Monitoring and Forecasting ... · Development of human resource and...
Pakistan: Satellite Based Crop Monitoring and Forecasting System
1
Ijaz Ahmad
General Manager
Pakistan Agriculture
2
Pakistan - Agriculture
SUPARCO Major Agriculture Areas
Population 190 million
Labor Force 54.90 million
Agriculture: Labor Force
26.1 million (44 .75) %
Crude Birth Rate
27.5 per 1000 persons
Crude Death Rate
7.3 per 1000 person
Annual Population Growth
2.05 percent
Pakistan: million ha
Geographic Area 79.61
Cropped Area 23.67
Cultivated Area 21.28
Cultivable waste 8.10
Irrigated Area 14.45 (67 %)
Forest Area 4.20
KHYBER
PAKHTUNKHWA (million Hectare)
Geographic Area 10.17
Cropped Area 1.68
Cultivable waste 1.23
Forest Area 1.32
PUNJAB (million Hectare)
Geographic Area 20.63
Cropped Area 16.96
Cultivable waste 1.56
Forest Area 0.49
SINDH (million Hectare)
Geographic Area 14.09
Cropped Area 3.83
Cultivable waste 1.38
Forest Area 1.03
BALOCHISTAN (million Hectare)
Geographic Area 34.72
Cropped Area 1.20
Cultivable waste 3.93
Forest Area 1.36
Land Utilization PAKISTAN Demography 2010-11 Rivers
SUPARCO
Sectoral Share in GDP (%)
20.9
25.8
53.3 Agriculture
Services
Monitoring of Crops through Satellite Technology
Pilot Project Coverage
FAO Mission (09 – 13 MAY 2005)
Stakeholders
UTF/PAK/101/PAK
Consultant Services
Trainings
Technical Audit of the technology
Crops Coverage (2013)
11
Wheat Cotton Rice Sugarcane
Maize Tobacco Potato
Crop Area Estimation
Area Frame Sampling System in Pakistan
13
• SUPARCO uses satellite based area frame sampling, a fully operational system for the estimation of crop areas, the system is in place since 2007.
• The Satellite Based Area Frame technique use a three stage stratification process to delineate homogenous areas in order to use statistical inference to estimate crop areas.
Strata Definition
11 Intense Cropland Area (75-100 % Cropland)
12 Less intense crop land (50-75 % Stratum)
21 Cropland Pasture Mixed (25-50 %)
42 Mostly Pasture ( 25 % cropland)
13 Un-identified seasonal vegetation
14 Areas rarely under vegetation
31 Rural area aroun11
d city (Less that 50 houses \ Km2 )
32 Inter city
50 Non farmland (Desert, Forest, Saline, establishments)
60 Water bodies (Rivers, Canals)
• Stratification makes it possible to produce more accurate estimates by reducing variability between samples.
• These statistical procedures are all part of a scientific probability survey system SUPARCO has implemented
Punjab South Zone broken down into PSU (red)
Design & Testing Sampling Strategies • To substantiate the satellite based image
classification system
• The spatial distribution of sampled segments as laid down by standard statistical/remote sensing techniques.
Area Frame System Sampled Segments
Sam
plin
g U
nit
Crop Calendar
• Improvement of data transmission of ground truth information through use of smart phones based on Android platform.
17
Ground Survey Data Collection System
Mobile Application Interface
Database management
17 Feb 2010 15 Mar 2010
Wheat
Tobacco
Satellite Image Classification
Satellite Image Classification
17 Feb 2010 15 Mar 2010 Enhanced Classification Classification
SUPARCO
20
Purana Dero (Jatoi Farm)
Satellite Image Classified Wheat
16-02-2010
Estimation Approach • Crop yield is a major component for the crop production forecast and estimation in
Pakistan.
• Statistical models are being used by SUPARCO for the estimation of the Yield. The following parameters are being incorporated on decadal basis in the model.
• SUPARCO currently starts with CRS yield estimates and adjusts them based on remote sensing and GIS technologies.
• Crop yield research at SUPARCO is currently focused on crop yield forecast modeling for early assessments and warning through monitoring of vegetation dynamics, agricultural inputs status.
• This yield modeling component is based on the modular approach of FAO of the United Nations.
Crop Yield Forecast/Estimates
Linear and Non Linear Yield Model
Developments
Satellite Based Vegetation Analysis
Weather Impacts on Crop
Irrigation Water Availability
Agriculture Mask Development
Crop Yield Assessments
Surveys
Fertilizer Availability
21
Crop Yield Forecasting
…Electromagnetic Spectrum (Light+)
…incoming light is
preferentially
absorbed (reflected)
depending on plant
physiology
Species
Photosynthesis
Water Content
[1]
[2]
[3]
[4]
Vegetation Image Behaviour
Crop Phenology
Dating of crop stages
Starting Dekad
Peak Dekad
Ending Dekad
Length of the Season
Length of Growth (Peak –
Staring Dekad)
Sum of NDVI value for different
time-periods
max_amp
Base line
onset_time
Growing
time
delta_amp
t
NDVI
t0
End of
platteau
Beginning of
platteau
Remote Sensing NDVI Image Processing
NDVI Image Analysis
Met. Station Data Spatial Interpolation
Agro-Climatic Analysis
Fertilizer Statistics
Irrigation Statistics
Crop Statistics
Area
Yield
Production
Decadal
Districts
Cro
p Y
ield
Mo
de
llin
g
Inte
grat
ion
an
d H
arm
on
izat
ion
Districts/Provincial
Monthly
Crop Yield Modelling
Crop Conditioning Approach Major crops are being continuously monitored by the PC1 programme across the country for crop forecasting, yield & production estimation to ensure food security
27-02-2011
31-03-2011
29-04-2011
25
Pakistan-Developing Crop Mask proposal Crop masking is a fundamental requirement in satellite remote sensing to conduct investigative research on crop forecasting /estimation of area distribution among all crops in both Kharif and Rabi seasons. Conducted study using both coarse and high resolution satellite data during peak crop growth season This endeavor (by FAO and USDA, executed by SUPARCO) will bring up the specific crop masks per province, designed for varying requirements rather than one crop mask under use for all seasons and crops. Objectives • Provide a mask showing distribution of different seasonal crops • Understand the pattern of different main crops in response to market condition • Extraction of crop statistics • Study of floods , developing flood extent maps and damage need assessment. • Assess the changes in agricultural land use • Database on spatio-temporal variation of different crops for decision making
Activities: Improving Sampling Strategies
Activities: Improving Sampling Strategies
27
A reliable and up-to-date land cover database has been just completed by
SUPARCO for Sindh and Punjab provinces. It provides:
• Description of the environment
• Information on land cover / use
• Baseline data on agricultural land Information to support the management of natural resources
• Information on human intervention on the land
The FAO methodologies and tools on land cover mapping were used to conduct the task, including one site trainings and backstopping missions by FAO experts in land cover databases production.
Land Cover of Pakistan
Interpretation keys
Training Workshop on Crop Information Portal and GeoNetwork software. 20-31 Jan 2014.
FAO and SUPARCO organized a two week training workshop in Islamabad:
• Jan 20-23: Administrating the Crop Information Portal
• Jan 24-27: Using the Crop Information Portal
• Jan 28-31: GeoNetwork opensource software
29
Nineteen experts from Crop Reporting Services of Punjab, Sindh, Khyber
Pakhtunkhwa and Baluchistan; from the Agriculture Universities of Tandojam
and Faisalabad and from SUPARCO
30
Future trainings:
• 04 x Area Frame
• 02 x Yield modeling using Remote Sensing and GIS
• 03 x IT system maintenance/operation
• 01 x Training on Exogenous Shocks
• 01 x Market Outlook Improvement
Distance Learning:
• AF development
• Acreage/yield estimation
• Utilization of auxiliary information on Remote Sensing
Main training topics:
• Crops Area Estimation
– Satellite based Area Frame Technique
– Satellite Data Image Processing
• Crop Yield Forecasting and Estimation
– Crops/Agri-Land Use Mask Development Techniques
– Crop Phenology and Growth Stage Monitoring
– Model Calibration matrices and analysis
Activities: Training Programs
• Project website • Crop Information Portal • Pakistan GLAM
Activities: Information dissemination and monitoring
31
http://dwms.fao.org/~test/home_en.asp
http://84.33.2.75/MapStore/
http://pekko.geog.umd.edu/glam/pakistan/
• University of Agriculture, Faisalabad (UAF)
The university's faculty comprises more than 500 teachers and accommodates over 5,000 students. University have a core staff educated in Remote sensing and GIS.
• Sindh Agriculture University, Tandojam
The university is an academic complex of five faculties, two centers, one constituent college and Directorate of Advanced Studies and Research.
Established activities Computer hardware , software and setup Development of human resource and technical capacities Post set-up consultations Establish fully functional units able to provide valuable agricultural crop statistics derived
from the integration of ground truth data and earth observation satellite systems Specialized short-term United States-based training (4 trainees) Technical training
Activities: Support to Universities
• Pakistan: Advanced Training on Monitoring of Crops through Satellite Technology
• Pakistan: How SUPARCO Makes Crop Forecasts and Estimates based on integral use of RS data
• Punjab CRS: Base Line
Survey
• Sindh CRS: Base Line Survey
Activities: Technical Consultations & Reports
Information Sharing
Monthly crop bulletin (Pak-SCMS)
Series of publications about rapid crop damage assessment (floods / droughts)
Technical reports
Crop Damage Assessment Reports
SUPARCO is publishing reports, beneficial for
disaster mitigation
Thank you
37